External Cheatsheets:
Numpy
-
Numpy matrixes are displayed as a list of row vectors.
- An identity matrix is therefore shown as
[[1, 0], [0, 1]]The first axis is the row number, then the column number. - axis=0; perform operations along the column, provides a row vector
- axis=1; perform operations along the row, provides a column vector (though it is still just a list)
- An identity matrix is therefore shown as
-
Initializing a matrix
# initialzing a matrix a = np.zeros(shape=(3, 2)) a = np.empty(shape=(3, 2)) #initializes with random values a = np.full(shape=(3,2), np.nan) -
Boolean array or matrix
a = np.full((m,n), True) b = np.full((m,n), False) -
string array with numpy1
my_array = numpy.empty([1, 2], dtype="S10") # S10 -> 10 characters # Add S_num to preallocate number of strings -
np.newaxis2
-
Numpy Array to list
np.array.tolist() -
max and max index
max_val = np.max(matrix) max_ind = np.argmax(matrix)
Pandas
-
Data import
data = pd.read_csv(os.path.join(root_folder, filename), sep='\t', skiprows=1, header=None, names=['col1','col2']) # to reset indexes (and drop it) data.reset_index(drop=True, inplace=True) -
Iteration3
# iterating through rows (NOT RECOMMENDED) for i, row in df.iterrows(): print(row) # iterating through index (probably not recommended) for i in df.index: print(i) -
Nested dictionary to dataframe (from_dict)
d = dict{1:{'a':0, 'b':1}, 2:{'a': 1, 'b':10}} df = pd.DataFrame.from_dict(d ,orient='index') -
Adding (extra) named indexes (rows) and sorting column based on named indexes
df.loc["mean"] = df.mean() df = df.sort_values("mean", axis=1, ascending=False)
Scipy
Interpolation
- https://docs.scipy.org/doc/scipy/tutorial/interpolate/smoothing_splines.html
- https://numpy.org/devdocs/reference/generated/numpy.interp.html#numpy.interp
Integration
JAX
PyMC
Setup
conda install numpy pandas scipy scikit-learn matplotlib seaborn dill
# optional
conda install ipykernel ipywidgets # for vscode + jupyter